Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network

  Advanced Search  

Incremental Diagnosis of DES by Satisfiability1 Alban Grastien and Anbulagan

Summary: Incremental Diagnosis of DES by Satisfiability1
Alban Grastien and Anbulagan
NICTA and Australian National University
Abstract. We propose a SAT-based algorithm for incremental di-
agnosis of discrete-event systems. The monotonicity is ensured by a
prediction window that uses the future observations to lead the cur-
rent diagnosis. Experiments stress the impact of parameters tuning
on the correctness and the efficiency of the approach.
1 Diagnosis by SAT
Diagnosis is the AI problem of determining whether a system is run-
ning correctly during a time window, and of identifying any failure
otherwise. Consider a system which is completely modeled by a DES
(basically a finite state machine) denoted Mod. This system is run-
ning and generates observations. The goal of the diagnosis is to de-
termine from the model and the observations whether faulty events
occurred on the system. The problem can be reduced to finding par-
ticular paths on the DES consistent with the observations [4]. Since
failures are rare events, we can consider paths that minimize the num-
ber of faults.
In [2], we proposed to solve the DES diagnosis problem with sat-


Source: Anbulagan, A. - National ICT Australia & Computer Sciences Laboratory, Australian National University


Collections: Computer Technologies and Information Sciences